Rufus Peabody: Prediction Markets and the Future of Sports Betting Analytics

Knowledge at Wharton

Ask about Knowledge at Wharton

Episode Description

ABOUT THE EPISODE Rufus Peabody, professional sports bettor, quantitative analyst and host of the @bettheprocess podcast, discusses the rise of prediction markets, comparing them to traditional sportsbooks while exploring liquidity, market-making, automation, and the evolving edge for sophisticated bettors. Plus, the Moneyball team has a wide-ranging sports analytics discussion covering Super Bowl matchup expectations, controversial coaching decisions, quarterback perception biases, and key storylines shaping the Australian Open.

ABOUT THE PODCAST Sports is a game of numbers. Wharton experts Eric Bradlow, Shane Jensen, Cade Massey, and Adi Wyner team up to tackle the world of sports, from current events to longstanding issues such as: What sports streaks are the most impressive? How do you rank the best players? Can athletes be compared across sports? Moneyball Highlights explains how decision-makers in the game can avoid the common mistakes and embrace the data. Episodes are recorded at the Wharton School.

#SportsBetting #SportsAnalytics #PredictionMarkets #SportsData #Sportsbook ----- The Wharton School of the University of Pennsylvania is committed to sharing its intellectual capital through Knowledge at Wharton, the school’s online business analysis journal.

Launched in 1999, Knowledge at Wharton offers free access to:

- Articles, podcasts, and videos highlighting Wharton faculty research and analysis of current business trends - Interviews with book authors from Wharton School Press - In-depth series of curated content like Wharton Executive Education’s Nano Tools for Leaders - A searchable database of more than 10,000 articles covering all aspects of business - A weekly newsletter that delivers Knowledge at Wharton insights directly to your inbox

Visit the Knowledge at Wharton homepage: https://whr.tn/3qUpSBE Subscribe to the Knowledge at Wharton YouTube channel: https://whr.tn/3DEAgAU

Read the full transcriptHide the full transcript
  1. Welcome, welcome to Wharton Moneyball. Welcome to a full hour of sports analytics here on the Wharton podcast
  2. network. This is Kade Massie hosting this week's show with the full crew. Audi Winer is here. Shane Jensen is
  3. here. Eric Bradlo is here. We're recording on Tuesday afternoon as we typically do these days. The show will
  4. go up on Wednesday. We are going to do a standard format. We're going to have a guest in the first half hour, then roll
  5. into open lines in the second sometime in the second half hour, we'll roll into open lines. Lots of things to talk
  6. about, but in particular, this Super Bowl week, or rather Super Bowl two weeks, the conference championships for
  7. this past weekend, we know who is on, and that means it's a big betting weekend.
  8. We've got a friend in the sports betting space, former guest, frequent guest in our early years in particular, Rufus
  9. Peabody, sports better extraordinaire, quant of some fame. Rufus, thanks for making
  10. time for us. Appreciate you coming on the show. >> Well, Kade, you taught me everything I
  11. know. So, it's >> I taught you everything you knew when you were like 20. You're like 40 now.
  12. So, I think I I think I can't claim that anymore. >> I'm a man. I'm 40. [laughter]
  13. >> You're almost You're almost a man. You're almost No, you are 40. You are 40. That's right.
  14. >> I know. It's It's crazy. I don't know where all the time went. >> Well, you spin it well. You've done very
  15. well. Rufus has been He went straight from his undergrad into Las Vegas. Spent about a year working for LBSC back in
  16. the day and then rolled into his own betting group really from one year out of school. Been doing that full-time
  17. since then. Rufus is coming to us from Las Vegas, one of the places he spends a fair bit of time. In previous years,
  18. especially early on, this would have been a manic week for Rufus. He's kind of famous for his prop bets, Super Bowl
  19. prop bets, but he's um he's his portfolio looks a little different these days, and he assured me he had capacity
  20. to do the show this week despite that. Rufus, we want to catch up with you in general. The whole crew enjoys time with
  21. you. Lots of things we could ask you about, but the thing we're most interested in hearing from you on is
  22. what's going on in the sports betting world with these prediction markets. People have been talking about it,
  23. especially over the last year. Khi Poly Market, the two big leaders burst into this space, kind of wrangle their way
  24. around regulations, state regulations that don't legalize sports betting in like California and Texas. They've blown
  25. up. They're being valued at billions of dollars themselves. Of course, sports betting is massive. FanDuel and
  26. DraftKings, these companies are cap market caps are like 15 billion, $30 billion in the case of FanDuel. So,
  27. massive market. We need to understand a little bit from you about what's going on here. So, you know, why is prediction
  28. markets advantageous from a better perspective in particular? Let me just say one last thing. Rufus has been an
  29. evangelist on the side of the better for years. his guy his cause is you know better markets better rules better
  30. access fair play in some sense for sports betterers and I think that's one of the reasons he likes prediction
  31. markets but let's hear from you Rufus first off Jade um they're it's predicting it's it's sports
  32. predicting not sports betting I think we should get that straight >> okay >> very very different
  33. >> very so different >> um but as you as you alluded to I I do think um one really good thing about
  34. prediction markets is that they are markets. There is a counterparty um [snorts] in the case of Kshi, it's
  35. anonymous that you know you nobody can sit the counterparty can't see that they're trading against you [snorts] and
  36. there's no cap to how much you can get down. Um well the cap is the market liquidity and so with sports books like
  37. FanDuel and DraftKings limiting sharper people uh it it becomes very difficult to be able to get down. So um so so
  38. that's a big advantage. The other thing is the you're going to in some cases have tighter margins especially if
  39. you're making markets rather than taking the fee structure the the fee structure
  40. at Khi right now is actually set up in a way that isn't super advantageous for taking. Um you're basically paying at a
  41. 50 cent a line at 50 cents you're paying like 1.44%. Um, but on top of the bid ass spread. So
  42. if something was trading like 50 51, you take it at 51, that's 52.44%. That's a little worse than minus one or
  43. sorry, you take it at 51. Yeah, it's a little bit worse than minus 110 there. But um but it's it it's so so in
  44. general, it's not super price advantageous for a recreational better. Um and especially on call she with the a
  45. lot of people uh are able to bet through um Robin Hood. They have a distribution
  46. through Robin Hood and other brokers and I believe Robin Hood also takes a fee. So that makes it really
  47. >> coffee. Really? Wow. Okay. Interesting. >> But overall I would say for for somebody
  48. like me who like let's say there's a game I make it 52%. I'm never going to be betting that in a sports book. But if
  49. I can sort of make a price at 50 cents or 49 cents, um, you know, I I I suddenly have opened up a door to a lot
  50. more games I can bet on. >> Does somebody see since Ruf I don't use that site. Does somebody see the size of
  51. the, you know, let's call it ask that you're putting out there because one could imagine that if a sharp is putting
  52. a million dollars out there, I'll just use some big massive number. um then someone might say hm more or less less
  53. likely to take that bet. So h how how do you think about that from a how you structure your play on prediction sites
  54. like this? >> That that is a great point and a great question Eric and and it is something
  55. that that everybody can see. So you you can you can't see you know you can see that there's three
  56. million shares bid at 51 cents on some game. You can't see that it's one person to bid 3 million for example.
  57. But so if if like something like the Super Bowl money line um I don't think you have to worry about that at all.
  58. It's going to be one cent wide and with massive volume on either side. The problem there is just getting your offer
  59. taken because you're going to be so far back in the queue. Um, but it is definitely a concern and we ran into
  60. this at some point, especially with um with some of these other sweep stakes sites we were doing predictions on. Um,
  61. like for example, Novig um that's a a prediction market as well. And >> say more about what a sweep a sweep
  62. stakes site. >> Uh sweep stakes technically you're not betting money. you are you're depositing
  63. money and you get like draft or not draft kings you get like no vig bucks or something like that
  64. I don't know what it's called but but but you're betting those things which can be redeemable one to one for dollars
  65. at the end so the way around >> their scheme actually >> yeah yeah it's it's yeah a lot of states
  66. are closing this sweep stake sweep stakes loophole uh though now >> okay >> but there there was a market there might be
  67. a market trading at um you know maybe minus 110 on each side or something like that with $500 and if you put up $10,000
  68. um the first time you know it it works out okay but then afterwards it's it people or or I shouldn't say the first
  69. time but I think for a little while people were able to get good fills that way. Um, let's say, you know, maybe I
  70. want I'll take the minus 110. I'll offer a plus 110 um to somebody else, right? And they can go arb that with FanDuel or
  71. DraftKings maybe. Um, but if you put up a big enough order, someone's like, "Oh,
  72. this is clearly somebody sharp and and you're going to blow up the market." >> Mhm.
  73. >> People are right. It's, you know, I might get a little bit of fill, but but I'm not going to get enough fill to
  74. justify blowing up the market like that. Mhm. >> But >> I will say prediction markets like why
  75. I'm one reason I'm enjoying them so much is because it is a different game and you do need to like like the game isn't
  76. just clicking a bet in. It's it is there are market dynamics. It's understanding
  77. the order book. Um it's trying to put a position you think is going to get filled without um without tipping people
  78. off. Like iceberg orders, right? Like we have things that'll auto like automatically re-trigger a new order if
  79. you will have an order. So so like you know but then which which is what you need in more
  80. illlquid markets. >> Mhm. >> But you also don't want to go so small that other people join you and then
  81. suddenly you you lose Q priority and can't down as much. So yeah there there's I mean I can talk about this for
  82. hours. So, but this this is the kind of thing that you used to never I mean you you were a price maker essentially. You
  83. were you were a model builder for years and this is more like the trading aspect
  84. of it. And there are some people like your uh your good buddy whose name escapes me at the moment who just makes
  85. a living like a very good living just doing technical trading stuff in the traditional sports betting markets and
  86. those are very different approaches but with the prediction markets it feels like you're moving a little bit more
  87. towards being intrigued by the technical aspects the trading aspects of the prediction market.
  88. >> Oh I'm I'm very intrigued by it. It's it's so fun like you know you can just explore these markets and learn
  89. something new every single day. like you know the I I've spent a lot of time trying to create a flowchart to
  90. understand the logic of some of these bots I'm up against and be like I can you know if I can predict exactly what
  91. they're going to do you know you can generally kind of figure out uh push up buttons and figure out like a weakness
  92. because every every bot has a weakness who's the most famous well not most famous but the most successful sports
  93. better in or not sport the most successful better in general I think in the history of the world. Um
  94. he says that the best trader will always will always be better than the best bot.
  95. I mean the goal is to every bot has a weakness >> and >> because they're too rigid because
  96. they're too predictable because it is algorithmic. >> There's logic. I mean they're trying to
  97. be human. You're trying to program in all the in all your intuition understanding markets perfectly and
  98. inevitably there's cases that come up that you haven't encountered before and a bot is going to behave in a very
  99. predictable manner. I think the best bots actually have some random should have some rand there's like a TTO bot
  100. right you know some of the time it's going to jump the order when it should you know some of the time it's going to
  101. join um >> you would think so right >> you put up 500 shares and it gets filled does it replenish right like maybe it
  102. does 95% of the time when it you know when you normally would right but you don't want to you you don't want to be
  103. too predictable and honestly it's the same thing for a sports book I think most sports book like sportsbook auto
  104. movers are too predictable. >> If if I bet into this line, I know they're going to move two points off of
  105. me. Um that that's really good information to have, but it also allows but knowing that allows me to manipulate
  106. uh uh or not manipulate, but like >> to to um to to move a line a little bit if I want to.
  107. >> Sure. Sure. Sure. Rufus, tell us something about your sense of the dynamics with the prediction markets
  108. coming in. How what percentage of your business do you send to prediction markets now versus traditional outlets?
  109. And what's your sense of other sharp better say like all the sharp money moved? What fraction of the sharp money?
  110. It's an absurd question, but what's your sense >> have? I think it's it's still a small
  111. percentage for most people. I would say relatively. I mean to I guess to give you some perspective we've done in the
  112. last 4 months or so I think something like 165 170 million contracts on KHI but the other thing with KI is it's less
  113. predictable so each contract has to win a dollar um it it's less predictable though so
  114. there's some games where we will get very big fills on things and others we won't and so uh something like the
  115. conference championship games, you know, I I end up with positions across a like
  116. a wide range of of outcomes like al alternate spreads, totals, um some some prop stuff, too. And I can't predict
  117. exactly exactly what it's going to come in on. And I I have to have, you know, well, I should say we cuz it's me and my
  118. team. We have our system set in place to have some some positional limits. But it
  119. gets kind of difficult too when you when you have all these sort of correlated things. You know, if if I get filled a
  120. lot on um Seattle minus4 and a half, getting the Rams minus 14 and a half doesn't is that's not a that doesn't
  121. really offset it that that much, right? Um but so so there's all these Yeah. So So there's I'm not answering your
  122. question here, though. I I would say >> No, no. One thing you're saying is that it's it's less than I would I might have
  123. thought. The way you talk about the benefits and the value, you might think that you put all your business there,
  124. but you're saying there's still limits to what you can do. >> Yeah. I mean, it's if we want to fill
  125. $50,000 on a college basketball side, like we're still not generally doing that on a prediction market.
  126. It's, you know, we don't have the certainty we're going to get down that much. Like it's there's certain markets
  127. for example like the conference championship games NFL games where you can get tremendous
  128. amounts of volume down very easily but others where you can't. You know most of these college basketball games like
  129. there's some that trade like $800 pregame total like across all like spread total money line you know if it's
  130. like Drexel against Lyola or something. >> Yeah. Yeah. Okay, Rufus, what have been
  131. the sports books response to this? I know that DraftKings and FanDuel now are starting doing prediction markets of
  132. some kind, but you know that they have been the subject of withering criticism from people like yourselves. Are they
  133. learning? Are they adapting in any way or are they just going to play the same game and keep on doing their typical
  134. stuff on the sportsbook side? >> Well, I I I think that they see they see this as a risk, an existential risk, and
  135. probably rightfully so. And they do have their own sports prediction markets, I guess, of some sort. Like I know DRA
  136. DraftKings bets started and we looked into that. To be honest, I'm not convinced that DRA I'm I'm not convinced
  137. that it's actually a legit prediction market. I feel like it may just be DraftKings calling it that. Like part of
  138. me, the cynic in me says, right, because I I don't think it's an open I don't think it's like it's certainly not an
  139. open API and I don't think people can make on it. Um, you know, the cynic in me says DraftKings is just doing this to
  140. try to like bring down prediction markets so that they in general so that they go back to having their uh little
  141. little monopoly. But >> interesting. Okay. What does the industry think's going to happen with prediction markets?
  142. I know that it's turning in there's a regulatory question across states and even now federal government but there's
  143. also it kind of becomes a political question. Do people have senses of which way it's going to go? Well, I I think
  144. the consensus among the the sort of sharp betting community is that the next three years are probably going like
  145. prediction markets are a fairly safe bet especially given the amount of venture capital that's flowed into this and
  146. given the fact that uh Trump Junior's on the board of both Kali and Poly market.
  147. It seems it seems more than likely that unless the Supreme Court takes this up that that during the Trump
  148. administration prediction markets are probably pretty safe. um after that kind of all bets are off. So I think a lot of
  149. people were kind of focused on on that that sort of time frame. >> All right. Well, let me let Yeah, Audi,
  150. jump in, please. Oh, I'm sorry. Eric, I saw that there. Let me just Audi first. >> Audi, I've already asked a question.
  151. Aie, go and then I'll go after you. >> No, I'm just trying to get a lay of the land here. So, you're moving uh heavily
  152. into the prediction market. Um you never really bet in the FanDuel area in like because they limit you. Do you do sports
  153. bettings um more still in Vegas? Is that is that where things are shaking out for
  154. big bettors? >> We we sports bet out at anywhere and everywhere kind of it it's like um it
  155. it's this whole ecosystem Audi where there are people that there's like people that are movers that have
  156. operations that get accounts and can you know in essence get get like glorified middlemen, right? But but um like
  157. brokers if you will that exist. Um, we have we have our own accounts at at various places. There's the exchanges.
  158. We have people in Europe. Um, you know, we wish we had more connections in Asia to be honest because that's if you like
  159. China, that's where you really want to be if you can if you want to bet a lot of money. But, [snorts] um, it it's it's
  160. honestly it's a big logistical mess. Like I don't know how else to >> But that's no different than it has
  161. been. I mean, my sense of you, you used to be in a group of four four people and
  162. two of you guys were kind of the the number guys and two of the guys were more or less getting money down. Is that
  163. was that the right way to think about it? And that was kind of in traditional sports betting world. So, it's a a lot
  164. of the work has always been getting money down and it's just getting more and more complicated. It's
  165. extraordinary. >> Yeah. And it's it's it's very decentralized. We we use a lot of
  166. different people to help get us accounts to get to get down basically. >> Brad Loop.
  167. >> Yeah. So, let me ask you. So, just building off something you said earlier. So, since I'm Wharton's vice of AI and
  168. analytics, I might as well ask you an AI question. If I build an epsilon greedy reinforcement learning algorithm and
  169. ingest a lot of this data, am I going to start making some well, I'm going to choose an algorithm
  170. that um understands that if I make a bet today, there's some immediate payoff, but if I make a bet today, it's also
  171. potentially going to change the state of the world. If I, for example, could move
  172. the line, it could move people's perceptions of me. And so that's the way reinforcement learning algorithms work.
  173. Meaning there's an immediate payoff that I win this bet or not. Then there's it puts me it changes the future state of
  174. the world. And the epsilon greedy is what you said before which is I don't always necessarily choose the one that
  175. maximizes things. Sometimes I want to play a little randomness in there because randomness can have some
  176. benefits. So, have you seen in your world are people applying? Because now it's all the rage. If I was going to
  177. serve serve Rufus Peabuddy a display ad or if I was going to show you a price and I was a big tech firm, I would
  178. probably ingest a huge amount of data and optimize this using a standard computer science reinforcement learning
  179. algorithm. So I'm just wondering has that made its way as far as you know and the answer could be you don't know into
  180. the betting worlds >> in in in what sense in terms of modeling or in terms of bet
  181. like executing bets >> well someone trading off of that more on the quant side like it's telling you
  182. it's making some AI based recommendation of bets based on maximizing this objective function which is some blend
  183. of making money today and putting you in a better place tomorrow. I'm just wondering cuz this sounds to me like a
  184. very standard reinforcement learning problem, but I'm just wondering if like there's a ton of computer scientists
  185. just applying this stuff to data that they can scrape off these sites. >> Um, so in terms of like modeling actual
  186. events, like you know, I I think at some point like well I I'm not sure how much the future would
  187. be affected by how you model the event today. Um, I I think that I mean I think there's a lot of really smart computer
  188. science people looking at this stuff in general and especially um especially the
  189. the trading dynamics. I think that's where this is most interesting cuz how how you actually go about like trading
  190. like because you you have counterparties and let's say some counterparty has a weakness. If you exploit them like for
  191. everything right now, they're gonna wise up to it probably. But if you if you don't, you just take
  192. little bitty pieces and somebody else realizes this and then figures out the same weakness and then goes after them
  193. big, then you're like, well, I just left a bunch of money on the table. So I mean
  194. I personally I am not smart enough to to understand how how um how all the this this AI stuff works that you're talking
  195. about um the reinforcement learning stuff but um it's I mean it's such a like everything is changing so quickly
  196. and it's the kind of thing that I think is inherently very difficult to predict cuz you don't know I mean the
  197. market participants are changing right like you can build something and then this new company you know Jane Street
  198. launches um you know market making on on Khi or something and maybe it changes things right or or I mean look things
  199. are I I think we think of changes as happening more kind of smoothly but in my sports betting career I've realized
  200. that things kind of happen sometimes very suddenly like I thought I had this great bottle to to win betting second
  201. halves in college football in the NFL and I did for 5 years and then one year like I suddenly wasn't winning and and
  202. basically was just break even and I wasn't finding much value anymore. It's like the market had suddenly got more
  203. efficient. And it turns out that the guy setting the opening price at Bet Chris um who the the previous 5 years had
  204. retired um or or been told to step down and somebody new had come on who actually looked at the live lines before
  205. he posted the second half F line and and there were a lot of people out there that were holding that the weak openers
  206. that they had posted in place so that other books copied and then like people could get down and there was an
  207. opportunity But suddenly sharper openers had just happened just like that. Suddenly the opportunity was gone and I
  208. I kind of had no idea it was such a precarious like it was such like my edges were so fragile in that way.
  209. >> Right. Right. AI uh I got I guess I ask a question of a comment. I'll start with
  210. a comment because I like what you just said. I do with my sports and gaming analytics class an example. I look at
  211. old NFL lines going back like 65 years and I show that in every decade betting heavily on a big home [laughter] dog is
  212. a very profitable bet and then it and then all of a sudden it's not just and I get them all to bet on whether
  213. they should continue that everyone's like oh my god yeah it's great how we're not doing this and it just disappears
  214. one day and and I have no idea why. You might actually be able to tell me why but it just is gone. Um so that's my
  215. comment. My question is guess sort of related to what Eric had to say. um when you're looking for edges, I can imagine
  216. there's two directions. One is to just build better models and have better prices and know the probabilities better
  217. and find mismatches and buy them. Um, the other would be to kind of look at just in general at the behaviors of the
  218. markets and find out there's anything in the betting the the the the demand structure the way and this is kind of
  219. what Eric was talking about where you can find bets not by knowing not by studying say football or baseball but by
  220. studying the market and figuring out if there's advantages because I think that the question is Eric has is about that
  221. and you're like how is reinforcement going to learn about football you know and which it's not by the there's no end
  222. effort to do that something else and I imagine I don't know these markets well there's probably opportunities in both
  223. areas but but I but how about in the in the way this the bets are structured I I
  224. mean I think that's the whole I mean that like like so we're building automated market making stuff and this
  225. is what I've spent the last like four or five months like primarily doing um aside from my normal you know I still
  226. have to run my golf model every week run the Massie Pbody ratings etc. But but the whole idea of understanding the
  227. market dynamics, understanding like um how the market moves and I mean I I trying to trade and get two-way action.
  228. And what I've learned is that a lot of there's a lot of bots out there and by bots I mean automated market makers that
  229. other people are, you know, have created um that don't know anything about the mark like the the the sports betting
  230. market as a whole and are just trading based on on the trades of other or the orders put up by other market
  231. participants. And I've found some that are very weak at this because they have nothing to
  232. anchor to. um you know in some wider markets they can be uh pushed to a to give you a really good price. Um you
  233. know there there's bots that literally in illlquid markets you know let's say something is trading um maybe it's like
  234. a no touchdown market. It's it's trading like 82 cent bid 97 ask. Um you know if
  235. I put up 85 they'll put up 86. So I put up 87 they'll pull up 88. if I put up, you know, 93, they'll put up 94, right?
  236. I mean, there's there's there's some bots that are just like infuriating in in certain ways. And so the ones that
  237. are infuriating, you kind of want to teach them a lesson so they stop so so they, you know, but I I I guess just to
  238. kind of comment on this, I feel like I maybe brought up that like there's this like there's really two dimensions here.
  239. There's the predictability of an event and then there's the irrationality of behavior around that event. You know
  240. what I'm saying? And and and I kind of feel like I guess yeah, Audi was talking about like you can build better and
  241. better models for finding out which events are truly like unpredictable versus not or getting an edge, but it's
  242. like again without without even a comment on how predictable the actual event is, are there certain behaviors?
  243. Yeah. like either bots or people exhibit that kind of like are are you know arbitrageable or whatever like it's kind
  244. of an irrationality of betting behavior whether it even automated or >> okay no no like no it was more of a
  245. >> but what I think one of the things >> for me at least it increased my instinct
  246. to kind of have it on these two different dimensions >> I will say you're right like so as Kade
  247. said like I've spent most of my career building predictive models and and being less focused on the actual trading and
  248. and I've had partners doing that, but that's more logistics than anything else. It's it's not like, you know, we
  249. understand market dynamics, but the market like in general is going to move towards our price and we want to get
  250. down as early as we can where we can still get enough money down, right? And and prediction markets are very
  251. different. Um there is this whole it's a whole another game you're playing. Um not trying to tip your hand. um you know
  252. like let's say I want to bet um and and actually putting up orders at prices that you don't even think are
  253. good because if you know you can get two-way fill um you can effectively get a better price that way. So, let's say
  254. the most I want on a particular event is 200,000 contracts. Um, and let's say I price this event at 60%. 60 cents. You
  255. know, I could get let, you know, I could get 200,000 contracts at 55 cents and that's a great bet. But what if I could
  256. get a million at 55 cents and then sell 800,000 at 57? That's way better, right?
  257. And so a lot of it is like predicting where the order flow is going to come from. And like I made up this word fill
  258. velocity where I you're kind of like it's my idea of like trying to figure out um and I haven't actually built a
  259. model to do this but but it's fascinating to me because I don't it's very I I don't think one can do this
  260. without AI by the way but basically given how how quickly something's been filling what the top uh you know how how
  261. deep the queue is how many contracts are at the sort of top ask price. So, um, and like trying to figure out what the
  262. likelihood I get filled in, you know, a certain amount of time is. Um, but then,
  263. you know, you can have this and then something changes. the market move as a whole moves and and so it's you know the
  264. way Kali is set up with 1 cent tick sizes um is m makes it a very different game because it is so Q position uh or Q
  265. priority is such is so important. So you really have to sort of like think ahead
  266. um and and try to like like if I want to get filled on both sides of something and you know where one side might not be
  267. posit isn't positive EV for me but overall I think I can get a better price that way like I have to put that order
  268. up before I even get filled for the first you know the side I actually want right and so I mean there's some games
  269. but but you know there's nothing more satis satisfying to me than some game where I was able to like,
  270. >> you know, take one side at 85 cents and the other side at like 12 cents. It's
  271. just even even though it's not the most it's not those aren't the biggest EV trades I'll make, um there's something
  272. satisfying about getting filled both ways and especially if I have something up and I literally see the action coming
  273. in on both sides, I'm kind of like, you know, it's it's basically what a sports book is, you know, trying to do. Are are
  274. you saying they're happy to get arbitragees, but they're not that they're not they're not really that much
  275. profitable because the the gap is tiny, >> right? I mean, well, I'm I mean, they
  276. can be quite profitable and the gaps, you know, aren't the gaps are, you know, you won't see a market sitting there
  277. with like a, you know, 90% over round or something like that most of the time. But, um, normally it's going to add up
  278. to like 98 or 99 cents. Um, >> Rufus, one of the things I think I'm hearing from you is that you you and I
  279. think maybe I've heard from you this other places is that the combination of understanding
  280. fundamental value and understanding market mechanisms and market dynamics. That combination is especially powerful.
  281. I think that combination is what sets us apart because I think a lot of just market makers don't have something to
  282. anchor to or they're anchoring to the market the sports betting market as a whole but
  283. the sports betting market as a whole is also often times like well there are large moves sometimes and and it there's
  284. times when it's more fragile than others and so um you know somebody can be like
  285. oh the market's minus 2 and a half like I'm very happy to just take collect, you
  286. know, to to give people big bets, you know, if I'm if if if I'm getting 47 cents on the dollar, basically if I'm
  287. getting the plus 110 or plus 112. Um, but that market might, you know, not actually be the most efficient thing at
  288. the moment, especially if there's more volume on Koshi. I think that's the bigger thing. There's there's we I've
  289. seen this multiple occasions where um I'll get hit and then after that somebody will hit the sports books that
  290. uh the on-screen sports books and basically the the sports betting market, DraftKings, FanDuel, Pinnacle, all those
  291. books will move afterwards because normally you think like the thing is most market makers are sort of using
  292. this the market the sports betting market as a whole as this sort a sense of truth but or
  293. yeah that is that's their sort of golden or their northern north star I guess but I mean it's I I it's it's
  294. interesting to see if how it'll keep changing because I think a lot of you know what makes something efficient
  295. right price discovery where is there more price discovery being done you know in some cases it's
  296. on koshi >> right for sure okay Eric has a question But Eric, we need to wind this down
  297. pretty quickly. So, what do you got? >> I was just going to ask one last question. Like, how complicated can bets
  298. be in the sense of is it ever like the way we're going to make money on this, let's call it event is to make 17
  299. different bets, like three of them on FanDuel, four of them contracts on Khi, six of them through this or that's just
  300. over complicating things. It's never I mean I'm not saying you couldn't do that but is that just not ever done?
  301. >> I mean I it sounds over like it's over complicating a little bit but I think it
  302. all depends on price. If you can get the best price on FanDuel you know then take
  303. the FanDuel. If you can get the best price on KI take the >> KHI. This is a a lesson you hear from
  304. Sharps again and again and again. It's like price shop. Find you you need information go to the places where the
  305. prices are best. That's going to be the quickest way to get yourself in an advantageous position. Mhm.
  306. >> Um, Rufus, thanks for all of that phenomenal, very dynamic moment in that whole world. Before you go though, this
  307. thing I've said about you for years is fun to say. I say you're the world's best golf better. And I and it's fun
  308. because it's a ridiculous thing to say, but it also might be true. And um, so before you go, give us something on
  309. golf. Scotty Sheffller wins his 20th PGA title this weekend. First, I think the second event of the year or whatever.
  310. Scotty Sheffer. Can >> it was his first event of the year, but not Yeah. >> Not the overall. Okay. Give us something
  311. on golf. Eric probably has some question about golf. Can you I know sometimes you're like, I'm not even paying
  312. attention to Scotty Sheffer. I'm just running my model and betting, you know, the 79th most likely guy because he the
  313. numbers are right. Do you have anything to say about Scotty Sheffler? I mean I think if you look at his last what two
  314. three years I mean it's right up there with it is close to peak Tiger. It really is. I mean Tiger Tiger sustained
  315. it for a decade. Um but but he really is on another level. I think it's kind of surprising because like Tiger looked so
  316. athletic and and his swing looks so perfect and and Sheffller looks like he's about to fall down half the time
  317. and and he he he doesn't really look like the athlete Tiger does but and it's not classically beautiful but it's you
  318. know he is you know what is it what do they say face to path impact or path path to face
  319. impact I don't know I mean he he's able to consistently do it and and his distance control is
  320. just on another level. Um, and I think that's something that is I think has separated him uh more than anything else
  321. is his distance control with his irons is so much better than any other player on tour.
  322. >> Give us a sense in your golf modeling like what's the cutting edge for you right now? Like what are you trying to
  323. do in your golf model right now to make it better? What give us a sense of what the frontier is? Well, well, the
  324. frontier is doing things that other people aren't doing, Kade, and other people haven't even thought about doing.
  325. And and so if I actually talk about it, then it it becomes not as valuable. [laughter]
  326. >> Give us something you did a year ago that was that was on the frontier and now is no longer on the frontier. What's
  327. an example of something you did you thought maybe [snorts] nobody else was doing even five years ago?
  328. Well, I mean, there's so many interesting things I want to talk about here and I would if this was private and
  329. but I would say look, I think it's just it's just asking interesting questions and
  330. trying to quantify more nuanced elements of skill. >> Okay. Okay. Well, that's that's actually
  331. informative that you're you're talking about going pretty obscure places to find edges essentially because all the
  332. basics are baked in. Does it have to be baked on top of an already very sound structural that's got the structure
  333. that's going to capture most of the fundamentals or can you come in kind of new if you if you've got a new thing and
  334. it's the only thing you got is there edge there or does it have to be on top of a foundation? For me, it's on top of
  335. a foundation. But I guess it depends on if you're like if you could come in and say, "Okay, the market is broadly
  336. efficient, but I think I found that this certain thing is undervalued or overvalued or I I don't think the market
  337. is fully capturing this." Um, you know, that that's certainly one one way you could could do it.
  338. >> But I'm not using the I mean, I'm basically making the market. >> You use the market as your foundation
  339. basically. All right. You got your model. Yeah, your model's just getting sounder and sounder. Okay.
  340. Um, Rufus, we'll let you go. Thank you for the time. I do want to say that Rufus is the co-host of a longunning
  341. sports betting podcast. It must be the best established sports betting podcast out there, Bet the Process. Bet the
  342. Process with Jeff Ma. Rufus and Jeff have been doing that for a long time. They, you know, they knocked around for
  343. a couple years and they got real serious about it. It's a pretty serious show. It's a fantastic show. Strong recommend.
  344. Bet the process. You can catch Rufus and Jeff there most weeks. Rufus Peabody, track him down. Catch him anytime you
  345. can. Always entertaining. Rufus, thanks for the time. Love seeing you. >> Hey, thanks so so much for for putting
  346. up with me and and all my digressions on predition markets. >> We barely got any digressions by Rufus
  347. standards. That was a linear conversation. That was real tight. Thank you, Rufus. That is the first half of
  348. Wharton Moneyball. We will have a second half. Come back and join us after the break.
  349. Welcome back. Welcome back to Wharton Moneyball. Welcome to the second half, second part of the show. We ran a little
  350. along with our guest, Rufus Peabody. Longtime friend of the show, my collaborator on Massie Peabody for many
  351. years now. Rufus graduated almost 20 years ago. We're coming up on I think this is 18 would be 18. Been in the
  352. industry for a long time. Informative discussion with him about prediction markets in particular. Fellas, Super
  353. Bowl 60 got set on Sunday. Unsurprisingly, the Patriots made it past a quarterback list Denver Broncos
  354. in the Blizzard and they're going to be in their 47th Super Bowl. Did you all know the
  355. Patriots have been in 47 out of the 60 Super Bowls? It's really [laughter] high number.
  356. >> They are playing, who are they playing against? The Seahawks. The Seahawks beat
  357. the Rams. the a the NFC champion NFC West division championship that produced the Super Bowl champion here. The
  358. Seahawks are favored early lines holding tied at four and a half I think. Um this
  359. game is in San Jose, sometimes referred to as San Francisco. Um Shane, God almighty, I mean what's it like to be
  360. from New England these [laughter] days? No, I mean to I I mean I'm not I'm not actually from New England, but to have
  361. arrived there on the scene in like, you know, 1999, it was very >> timing timing was good, Shane.
  362. >> Timing was good. >> Yeah. No. And uh this this is probably the most unexpected of all of them. I I
  363. I think thus far this run, um I think it's been a I think they're an intriguing team because I think we could
  364. still I don't know what you guys feel, but you can convince yourself they're not actually even that good. [laughter]
  365. wanted to uh but they're in the Super Bowl so I don't I don't know. Um and I think that you know if if anything I
  366. think if I was setting the line I mean sharper minds than I do set those lines but I if I was setting the line it would
  367. be even even more substantial. I mean just based on how good Seattle looked um against the 49ers. I think they're
  368. really kind I mean the only hope I guess the Patriots have is that time Donald sees Ghost again or whatever.
  369. >> Yeah. >> Well, what other reactions from the weekend? Um the the the Ram Seahawks game was tight as
  370. as as would be expected. Fun could have gone either way. The Denver New England top game was a lot tighter than a lot of
  371. people would have thought and that snow added wonderful flavor to it. Um Payton forewent that field goal opportunity
  372. early. That sure did feel painful pretty quickly when nobody else could score. Um
  373. any other observations coming out of the championship round? >> How do you guys feel actually about that
  374. decision? Like let's let's assume full information like Sean you're Shawn Payeyton and you have the weather
  375. report. Maybe you don't know it's going to go quite that dramatically like like like go to unplayable levels but you
  376. know the it's going to get worse. I feel like that maybe kind of I feel like the
  377. analytics probably suggest that you shouldn't you should go for you know on fourth and one you should go for it
  378. there. Does you know if you can incorporate that kind of you know future weather does that move the needle
  379. enough? >> Weather was part of it but also just you know what are you forecasting about your
  380. offense's ability and stood them and I what I I think the most interesting bit is that they had they drove down the
  381. first um drive and possession and then the second they go down again. So they've
  382. got they've had the ball twice. They've moved the ball really well twice and he probably wasn't thinking I'm going to
  383. get 35 more yards for the rest of the game. you know, that's that's the part that really fell out of sync with the
  384. first part of the experience. So, I'm sympathetic to him in that way. >> Yeah. I mean, the reaction I had at that
  385. play wasn't so much I like the way Shane framed it. If I knew the weather report
  386. was going to be bad, I might have changed. But also, I I'll go look, I'll go to my deathbed
  387. by saying I don't understand that play call. And let me say why I stayed out. >> Oh, really? Oh, wow. Well, maybe you're
  388. going to catch them off guard, but again, you're trusting a crucial point in the game to a quarterback that has
  389. not played in over two years. Why are you throwing the football? >> Oh, the play call. You mean the play
  390. call? I don't going for it. Okay, that makes more sense. >> Not that call. That's what I don't And
  391. that's how Shawn Payton and that's why what I put in the rundown. Did Shawn Payeyton maybe, and I love the way you
  392. said it, Cade, because of the first two series, actually it was the third series
  393. because the first series they were three and out, but the first the second and third series, maybe he saw Stidum doing
  394. so well, he said, you know what, I'm going to trust this guy on this crucial play to throw the football. And to me,
  395. it was the right call to go for it, but not that play. No. And also to me, I've got, look, I've said this about the
  396. previous Seattle Patriots Super Bowl, too. You say something to your team when you don't think you can run the ball one
  397. yard on fourth down. You don't think you can do cuz if you thought you could, you
  398. would do it. And this idea that I'm going to out trick them. It reminded me, it did. reminded me of Pete Carol having
  399. Russell Wilson throw the football when you got beast mode running the goddamn ball and I think even on the Patriot
  400. game, maybe you can correct me, that wasn't even fourth down. That might have been first or second down.
  401. >> I mean, it was it was it was in fact second down though. I will note the first down was them handing it to beast
  402. mode and him not getting into the end zone. >> I understand that. That's all I'm
  403. commenting on where they >> tried to do exactly. It reminded me of >> You don't have to rehash that history.
  404. It just reminded me of the coach trying to almost be too smart there. I liked going for it. I didn't like the play
  405. call. >> And I and I you you're you're kind of echoing. My general frustration with
  406. fourth down decisions is after the fact we analyze them. And so far so so often we're we're convolving the decision to
  407. go for it or not with like a really suboptimal play call. Like I mean my my bugaboo is always like you know some
  408. kind of fourth and three and they're in shotgun or something like that. fourth and one and they're in shotgun. You
  409. know, the these types of kind of, you know, I wish we could also observe the be what what happens in the alternate
  410. universe where they kind of did the optimal thing at least conditional on that. >> But let's also let's talk about
  411. something Shane said earlier did a better thing. >> Suppose I told you that let's imagine
  412. they had kicked a field goal there, okay? And let's say they have 10 points. As a matter of fact, let's even imagine
  413. that was all they scored in the first half cuz people are talking about the Patriots defense and it is a good
  414. defense. But let's not make it seem great. Let's just remember >> Jared Stum who hadn't thrown a pass in
  415. over two years. Let's imagine he put up 10 points in the first half against the Patriots. That's not the 85 Bears
  416. performance. That's my point is that like nothing. So right now, if I had to update any of the
  417. >> Patriots didn't score 10 points in that Super Bowl, [laughter] >> right? Well, no, no, but but also if you
  418. had me update any part of like after watching the two games, if you had me update anything, here would be my
  419. update. If I think about the four sides of the ball, I was disappointed in the Patriot offense. I think the Patriot
  420. offense is worse than I thought going into the game. And I think Seattle's offense is better than I thought going
  421. into >> in I can't disagree generally, but in evaluating your in your evaluation of
  422. the Patriots offense, are you building in that they have faced three of the top five defenses now in their playoff run
  423. in the NFL? because they have the Denver. Denver, Houston, and San Diego all have I mean
  424. San Diego's five, but or sorry, LA Chargers are number five, but like yeah, those are three of the top five defenses
  425. and as well as snow conditions. And again, I'm not I'm I'm playing devil's advocate because I'm actually on your
  426. side on this one. It also I I I also came away underwhelmed with the uh Patriots offensive performance. But this
  427. is kind of, you know, the counterargument that people are making right now is they have faced some very
  428. legitimate, you know, kind of, you know, maybe not 85 Bears level, but strong defenses in this run and have done
  429. barely enough to win and been lucky while they've been doing that. But, >> you know, this is this is why it's fun.
  430. This is why we're going to play the game. >> I will say Seattle is another great
  431. defense. So, it's not like if you get to be better or anything like that. >> My our friend and colleague Joe Simmons
  432. uh wrote uh over the weekend about comparing the media's reaction to Drake May versus Sam Darnold. And
  433. >> the media versus who >> on on May versus Darnold. And you know, Darnold like anytime something goes
  434. well, you know, it's probably the receiver or the offense or the offense. But with anything goes wrong, it's like
  435. Darnold because he's a flawed guy. And then May is like this darling. And is their performance really that different?
  436. I mean, can we really make these kinds of And do people Here's the thing. Here's the premise, I guess, is that
  437. guys get these reputations early and that they color for years. They color people's impressions. And I feel like I
  438. mean, look, I'm not I don't pay enough attention to know in detail, and some people pay a lot of attention.
  439. >> But I love that Darnell is doing as well as he is, and I feel like he's just born
  440. through bad early experiences in a way that most don't. But I don't doubt that that negative early experience is still
  441. coloring people's perceptions of him. >> No. And I I mean I think it's amazing. Such a wonderful career arc for him. Uh
  442. I mean if they weren't playing the Patriots, I'd be probably cheering pretty hard for him honestly in the
  443. Super Bowl. But uh but yeah, question Shane, who would you rather have as your quarterback for the Super Bowl right
  444. now, Drake May or Sam Darnold? >> Great question. >> Great question. >> Drake May. Drake May has never seen
  445. >> Sam Darnold can be had. We've seen Sam Darnold be had. He has not been had in
  446. this run. He has not been had this season, but as recently as last season's playoffs, he can be had. I'm not saying
  447. they necessarily will get him, right? But >> I'm going to take the guy that's won,
  448. including the playoffs, I'm going to take the guy that's won 31 games the last two seasons or whatever the number
  449. is. But the only guy besides Tom Brady, >> you would have taken Kurt Warner going
  450. into that super first Super Bowl, too, right? Come on. >> Yeah, I would have. >> Oh, for sure. But it's an interesting
  451. question. I like this with Eric. I'm going I'm going Darnold as well, but I realize it's an interesting question and
  452. I am I am worried about, you know, next next level stakes, >> how things how things go for him. Um,
  453. all right. What else do we have around the world of sports? Why don't we give Eric a little time on the Australian
  454. Open? the first major of the eight majors we'll see in golf and tennis. The first major is being played right now
  455. down in Australia. >> Yeah. So, I mean, let's ignore last night for just one second or the the
  456. matches that would happen. Something I put in the rundown. So, in the final eight here were the men's seeds left. 1
  457. 2 3 4 5 6 8 and 25. in the women's 1 2 3 4 5 6 12 and 29. >> Now that's remarkable.
  458. >> Yeah. >> The amount of top seeds that are still in. Now I will say last night um 29 got
  459. eliminated badly by one. Sabalanka uh beat um I forget the woman's name. The American woman that uh I forget her
  460. name, but she got dusted off pretty quickly. >> 29th, >> what' it say? >> She was 29th ranked.
  461. >> 29th ranked. Um 12vidilina destroyed Koko Goff yesterday actually. So 12 is in the semi-finals. Matter of
  462. fact, 12 is playing one. Sidilena is playing Sabalanka in one semi-final on the women's side. on the men's side.
  463. Wow. I mean, Alcarez played brilliantly. I mean, he played the number six in the
  464. world, beat him 75, should have been 63 in the first set. 62 61. And the guy's Australian, so playing on his home
  465. court. um Alex Demenor who had I think he was 15-1 in his last 16 matches and >> got pedestrianed if you'd like by um by
  466. Alcarz. The interesting part will be you know Jookovic. So here's the here's the
  467. question. He got defaulted into the quarterfinals. So he has only played three matches not four. What does it
  468. mean got defaulted into >> the guy he was going to play in the quarter in the sorry in the round of 16
  469. defaulted before the match. >> He was going to play the number 16 seat who had beaten him the last time they
  470. played by the way Jacob >> Mensik. >> Okay. >> He didn't have to play him. So he's had
  471. four days rest now. >> His first three matches the top ranked player he played was number 85 in the
  472. world. He's played two wild cards in an 85. He beat them all in straight sets. So, he's in the quarterfinals and he's
  473. played nine sets of tennis so far. >> He's got to be fully rested. >> Now, he's playing the number six or five
  474. in the world, Musetti, who I think he's not. I know he's nine and one against. Now, here comes the major question.
  475. What's going to happen in the next round potentially when he plays S? That's the
  476. magic question. Can he on S side of the draw, so he'd have to get past center to
  477. get to the finals. But this I my question coming into this conversation was to explain to me center right now.
  478. So there was this match sometime this past week where if not for the heat regulations
  479. >> I can tell you a personal story. >> So but so the question though that I want you to land on is what's your
  480. assessment of center's condition right now and why is it any different than the top level it has been?
  481. >> Yeah. So just so you know it was like 105 degrees out. The reason I know that
  482. match well is because the person he was playing against is the twin brother of a
  483. guy who was on Zack who's on the pen squash team. >> So Nick Spazeri was number one and two
  484. for about the last five years on pen squash team. His twin brother is Elliot Spazeri who's the one that S plays. So I
  485. was watching the match. >> Um yeah, S would have lost that match if they didn't close the roof and they
  486. didn't have a 10-minute break after the third set. He was cramping severely. This is something that has happened. The
  487. thing now is is that they're now down to so few players. They're going to be playing all the matches in Rod Labor
  488. Arena with a shut roof cuz it's 110° out. So, it's going to be 70° inside. So, um the no there's nobody on this
  489. planet more upset that they shut that roof in the middle of that match than NovakJokovic because he saw center going
  490. out. Now, his path to the finals is clear. And just to be clear, last year, last year, not five years ago,
  491. >> he beat Alcarez in the Australian Open in the quarterfinals. So he was thinking
  492. he's one inch away from potentially steamrolling into the finals against a guy he beat.
  493. >> Okay. >> On the on on this surface. >> Yeah. Yeah. Yeah. >> But no, I don't think there's any
  494. lasting uh Center looked great in his last match. >> Um I believe, by the way, Center has a
  495. very intriguing match. It'll be at like 3:00 or 4 in the morning. I'm pretty sure he's playing Ben Shelton. Now,
  496. that's no gimme a match. I mean, that's a very interesting match. >> Tell us more about Ben Shelton.
  497. >> Well, Ben Shelton's been rising up uh quickly. He was a collegiate squash play
  498. uh tennis player. Um he's got about 140 mph serve and so when he wants to, if he
  499. gets it in, he'll just blow you off the court. extremely athletic, great serving
  500. volier, allcourt tennis guy. Um, a lot of people believe if there is a third player that will be joining the big two
  501. of SR and Alcarz, it will be uh Ben Shelton. >> And so, um, do I think he's going to
  502. beat S, the two-time defending champion at the Australian Open? No. Do I think he's going to beat the guy who I don't
  503. think he's lost in maybe 30 consecutive matches on hard courts? No, I do not think he's going to. But it's not a
  504. trivial match. And so look, it's all it's going to come down to it's it's a fascinating f And by the way, do not
  505. assume that Alcarez is making the finals. Alcarz is playing Zarev in the other semiinal. And by the way, their
  506. career record against each other, six and six. >> So this this idea that it's a feta
  507. comple like bet the house that it's center versus Algarz. I ain't betting the house on that. that ain't I mean
  508. there's no I mean look Al Perez is six and six against Verv so there's no guarantee he'll be the
  509. favorite but not that heavy a favorite and center still has to get past Ben Shelton and probably Novak Djokovic I'm
  510. not saying he doesn't have a 60 or 70 or 80% chance to beat them both but it's not for sure it's a fascinating both the
  511. men's and the women's side is extraordinarily exciting this year >> well strong draw everyone's still in a
  512. lot of LA fun last few days here as we roll into the finals on on Sunday. Um why
  513. don't we wrap it there? We just lost Audi and we've run about our full time now. So we'll wrap it there. We will
  514. come back next time of course. Until then, thanks for Audi Winer and Absentia on behalf of Eric Bradlo and Shane
  515. Jensen, Kate Massie. Many thanks to Dion Simpkins who makes the whole thing happen. Marissa Raina, our producer and
  516. booker, very helpful. the big boss lady D Patel. Many thanks to D and to you guys for listening. Come back and join
  517. us next time. Between now and then, enjoy your sports.

Brief Summary

Rufus Peabody discusses prediction markets, Super Bowl betting, and golf analytics on Wharton Moneyball.

Key points

  • Super Bowl Betting Insights. Rufus Peabody shares insights on sports betting, particularly around Super Bowl prop bets.
  • The Rise of Prediction Markets. Discussion on the advantages of prediction markets over traditional sports betting.
  • The Impact of AI on Betting. Exploring how AI and reinforcement learning are changing the betting landscape.
  • Market Dynamics and Predictability. Understanding how market dynamics affect betting strategies and opportunities.
  • The Fragility of Betting Edges. Discussing how quickly betting opportunities can disappear due to market efficiency.
  • Bet the Process Podcast. Rufus Peabody co-hosts a serious sports betting podcast that has gained popularity over the years.

Episode highlights

Tension

45/ 100Some tension

Related episodes